[Freeciv-Dev] Project goals - Multiplayer

[banjo <banjo@xxxxxxxxxx>]

Hi

I'm splitting the thread coz i it's discussing two different things...
multi-player & game balance. At least i want to add game balance to the 
discussion.  This post is about the multi-player issues, i'll also post 
about game balance.
Per Inge Mathisen wrote:
> The single player game suffers heavily under the restrictions imposed
> by multiplayer. The ban on modal dialogs has improved multiplayer
> significantly, but along with timeout and the undeclared ban on
> pausing the game to display information, it rules out many ways of
> displaying information to the player that can be highly beneficial
> for single player. Opening up a tab for tech selection and throwing it
> beneath the map tab is very nice for multiplayer, where it is
> imperative that you do not hide what happens in the early seconds of
> the new turn, and do not pause other players, but not optimal for
> single player, where it would not matter if the game waited for player
> confirmation of tech selection. Extending the rules of diplomacy for
> more interactivity with AI turned out to be extremely painful, and
> made me give up (see past threads).

The current (single player) turn model is based on rapid feedback to
the player, you give an order and the unit obeys. This is useful for
the player as it cuts down on the amount of memory and planning they
need to hold in their head.

In various multiplayer modes, this could let you be overwhelmed by a 
surprise attack, or demand that you have to be there for every move,
or make it awkward for the reading the dialogs etc.  This mode seems
to favour the fanatical gamer.

It is also completely historically unrealistic, for most of history
and even arguably today, the rulers got reports of events that were
in the past (pre telegraph they could be years old). The new orders
they wrote were not executed immediately, they first had to be sent
(by boat?), and then it took time for the orders to be executed.

What i am saying is that a write/execute method for handling orders
is more realistic, it may be easier to program, and seem to be what
is needed for multiplayer games.

It may be less desired for single player mode.  It may be harder to
code the ai, if it has to keep a record of what it has planned, as
opposed to just looking at what currently is.

Per Inge Mathisen wrote:
> Statistically, in multiplayer
> most game time takes place in the early game, since game are restarted
> often and many end rather quickly. Freeciv is IMHO quite boring in the
> early game, with few important decisions, few ways to surprise your
> opponents, and many repetitive actions.

In the ai.ruleset thread, Marko Lindqvist introduced the idea of being
able to script the ai grand strategy.  I dunno how complex it would be
to implement, but maybe it could help long term games. For example what
if the player could use ai scripting to develop ai policy, and if that
player misses a few turns, then the ai would carry out the policy.

Of course the ai would have to be slow to make changes, not starting a 
revolution and changing all the city builds just coz the player missed
a turn. It would add another level of game play if players could write
meta orders that were carried out by the ai.

Per Inge Mathisen wrote:
> Fixing multiplayer Freeciv would either involve a change to
> "turn-less" playing, similar to what Civ4 offers
> (http://www.civ3.com/devupdate_multi.cfm), which would not work very
> well for long-turn games, giving priority to shorter games, or a more
> radical redesign towards a Moo2 model where players queue up actions
> that are executed in the turn end, which is what will work best for
> long-turn games. (If at this point you ask "why not offer both as
> options?", you have simply not understood the depth of the problem and
> the amount of redesign required for either to work well.)

The first thing the link discusses is a ruleset that allows for rapid
games, more suitable coordinating multiple players.  This can be done
in the rulesets (by using a tournament ruleset), in the server config 
file (fast city growth, fast tech advancement) and by adding some new
victory conditions as options.

These in themselves are not major changes, and the first two are easy
to implement and probably already are.

Then the link discusses how turns could work, ive added more options
to this list, i think having all the possibilities laid out could be
useful....
* Turn Based
* Turnless
* Simultaneous Moves
* Play by Email
* Hotseat
* Queued Actions
* Real Time
* What we have now (is it Simultaneous Moves?)

Are there any more? Please add any that ive missed
Are there any meta patterns here?
Could someone define how each works?

Of all the multiplayer examples in that link, i like the play by email
one the best.  Especially when coupled with the idea of scripting your
ai to play for you when you can't.

Per Inge Mathisen wrote:
> Concerns from multiplayer often stand in opposition to making a better
> single player game. I think this opposition is our biggest problem. We
> can make a very good single player game, with the option of playing
> (turn-based) with 1-2 close friends. Or we can create a game that is
> fun in multiplayer. Because once we need to take into account the
> needs of higher amounts of players and a competitive and possibly
> hostile environment, the needs of single player must suffer.

In the few multiplayer games ive played they all were abandoned because 
of the length of time they were taking. I suspect this is a fundamental
problem.  Either have a long complex satisfying one player game ( which 
is the way i play freeciv) or have a simpler & faster multiplayer game.
I haven't tried long move freeciv, where i assume players get one turn
a day (with an online early game to get things started). But i imagine
there are issues there as well, in that an end turn goto attack, with
a early next turn follow up, may overwhelm a more casual player.

This is possibly realistic, (blitzkrieg, pearl_harbour), but i imagine
would annoy the victim, as they'd feel they didn't have a chance, this
is considered poor game design (ie not fun).

Per Inge Mathisen wrote:
> So perhaps what we need is a split of Freeciv into two projects, one
> for a multiplayer game and one as a single player game?

If that's what's required, then so be it. However, is it possible for 
there to be some backend part that's common between the two, with most 
of the forking being at the client end?  Or have i not understood the 
depth of the problem, and the amount of redesign required for this to 
work well?
-banjo